A statistical fallacy is a mistake in reading data rather than in the data itself — a number that is accurate, and an inference from it that is not.
Most of these are not fallacies in the strict sense — some are paradoxes, some are artifacts of how data gets collected, and one is simply a practice. What they share is that the arithmetic is right and the conclusion is wrong. Entries marked Fallacy or Bias live in their own library and open there.
Statistical fallacies are not errors of calculation. The sums are usually correct. The error sits in the step between the number and the claim — in what the data was never able to show, or in who is missing from it.
Some of these are properties of the mathematics itself. Simpson's paradox and regression to the mean are not mistakes at all; they are things that genuinely happen, and the fallacy is failing to expect them. Others, like survivorship bias, come from the shape of the sample rather than the sum.
They matter most where the stakes are highest: in medicine, in policy, in anything that reaches you as a headline with a percentage in it. A figure that is technically true and practically misleading is far harder to argue with than one that is simply wrong.
These sit alongside the broader logical fallacies and the cognitive biases that shape judgement before any data is collected at all.
